NOTES
Google dorking is structured search, not a vulnerability by itself. I use it during authorized reconnaissance to understand what a search engine has indexed for an in-scope organization: public documents, forgotten hostnames, login surfaces, old paths, exposed directory listings, and technology clues.
Every query starts with a domain I own or am explicitly allowed to assess. Search results can be stale, incomplete, or misleading, so I treat them as leads to validate inside scope—not proof of impact.
CORE OPERATORS
Limit the scope
site:example.com
Restricts results to the authorized domain. Use a leading minus to omit a noisy host or term.
site:example.com -www
site:example.com -"privacy policy"
Match a phrase or alternatives
site:example.com "exact phrase"
site:example.com (docs OR support)
Quotes keep a phrase together. Uppercase OR is useful when two terms describe the same lead.
Look at titles, URLs, and document types
site:example.com intitle:portal
site:example.com inurl:docs
site:example.com filetype:pdf
intitle: searches page titles, inurl: searches indexed URLs, and filetype: narrows results to a document format.
Bound results by date
site:example.com after:2025-01-01
site:example.com before:2026-01-01 after:2025-01-01
Date filters can reduce noise when looking for a recent migration, release, or documentation change. Indexed and published dates do not always mean the same thing.
QUERY RECIPES
Public documents
site:example.com (filetype:pdf OR filetype:docx OR filetype:xlsx)
site:example.com (filetype:pptx OR filetype:csv)
Authentication and administration surfaces
site:example.com (inurl:login OR inurl:signin OR inurl:portal)
site:example.com (intitle:admin OR intitle:dashboard)
Subdomains and non-default hosts
site:*.example.com -www
site:example.com -site:www.example.com
Directory listings and indexed configuration formats
site:example.com intitle:"index of"
site:example.com (filetype:xml OR filetype:json OR filetype:yml)
Finding a result does not authorize retrieving sensitive material. Capture the smallest evidence needed, avoid opening exposed secrets, and report the indexing problem through the agreed channel.
WORKFLOW
- Write the approved root domain and exclusions at the top of the notes.
- Start with a plain
site:query and record unusual hosts, paths, titles, and file types. - Add one operator at a time so it is clear which constraint changed the result set.
- Use safe placeholders in reusable notes; keep client names and sensitive results in the assessment workspace.
- Validate an interesting result against the live, authorized asset before treating it as current.
- Record the query, result URL, observation date, and why the result matters.
GOTCHAS
- Operators and result behavior change; a query returning nothing does not prove the content is absent.
- Search snippets and cached metadata can outlive the page that produced them.
- Do not automate Google result scraping or attempt to bypass rate limits and anti-abuse controls.
- The Google Hacking Database is a catalog of patterns, not a scope list. Rewrite every example around an authorized domain.
- Separate discovery from exploitation. Search-engine evidence is usually enough to raise a validation task.